International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 08 Issue: 09 | Sep 2021 www.irjet.net p-ISSN: 2395-0072 © 2021, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 765 AUTOMATIC IDENTIFICATION OF COVID-19 FROM CHEST X-RAY IMAGES USING ENHANCED MACHINE LEARNING TECHNIQUES Ankit Ghosh 1 , Purbita Kole 2 , Alok Kole 3 1 Student, School of Nuclear Studies and Application, Jadavpur University, Kolkata, India 2 Student, Department of Physics, Christ University, Bangalore, India 3 Professor, Department of Electrical Engineering, RCC Institute of Information Technology, Kolkata, India ---------------------------------------------------------------------***---------------------------------------------------------------------- Abstract - The Covid-19 pandemic has led to the loss of millions of human lives across the globe. The public health care system has been facing an unprecedented challenge since the outbreak of Covid-19. The accurate and the timely diagnosis of Covid-19 are extremely important. Artificial Intelligence (AI) and Machine Learning (ML) can play a crucial role in the fight of humanity against the Covid-19 pandemic. This paper presents a detailed overview of how different ML algorithms can be implemented for the automatic identification of Covid-19 infected patients using chest X-ray images. The dataset that has been used comprises of chest X- ray images of both Covid and non-Covid patients obtained from various sources. 9 ML algorithms namely, Support Vector Machine (SVM), Logistic Regression, K-Nearest Neighbour (KNN), Naïve Bayes (NB), Decision Tree (DT) Classifier, Random Forest Classifier, Stochastic Gradient Descent (SGD) Classifier, XGBoost Classifier and Gradient Boosting Classifier have been implemented to perform the classification of the images. A performance comparative analysis of the different ML algorithms has been conducted based on a few metrics such as accuracy, recall, precision, F1-score and the AUC-ROC curve. XGBoost has surpassed all the other classifiers with an accuracy as high as 93.9%, recall, precision and F1-score of 91.3% respectively and an AUC of 93.3%. Key Words: artificial intelligence, automatic identification, classification, comparative analysis, diagnosis, machine learning 1. INTRODUCTION Artificial Intelligence (AI) and Machine Learning (ML) have proven to be a valuable ally for medical practitioners [1] [2] [3]. Medical imaging data is a rich but complex source of information about the patient [4]. Although medical images like computed tomography (CT), magnetic resonance imaging (MRI), mammograms, X-Ray types such as fluoroscopy and angiography, and ultrasounds are a valuable collection of potentially life-saving data for healthcare researchers, providers and their patients, these kind of images present unique challenges that have traditionally limited radiologists’ effectiveness in the diagnosis and treatment of different ailments [5][6]. Medical image analysis is performed by radiologists and clinical doctors. Due to the increasing complexities of the images and overload of work the interpretations of medical images are at times prone to human error. There have been instances where one of four patients has experienced false positives from their medical image review. Undoubtedly traditional manual review and diagnosis of medical images have saved countless lives over the years. However, these methods can be improved. The ability of ML to analyse and learn from vast quantities of data has made them a favourite choice among medical and technology researchers [7] [8] [9] [10]. Image scanning using ML techniques have accomplished more reliable results. AI and ML also promise speed and consistency [11] [12]. Therefore, time sensitive cases can be addressed first and the diagnosis is also faster. AI and ML algorithms also eliminate errors introduced due to natural cognitive bias from clinical diagnoses [13] [14]. Therefore, by merging AI and ML techniques with the competency of medical practitioners, it is feasible to bring about significant advances in the field of medicine and healthcare [15] [16] [17]. 2. RELATED WORK A lot of research has been carried out in the field of AI and ML for medical image analysis. (Mamlook et al., 2020) have used a Convolutional Neural Network (CNN) model to perform the classification of chest X-ray images into health vs. sick for the detection of Pneumonia. Their proposed model has achieved an extremely high overall accuracy. (Sorić et al., 2020) have proposed the implementation of a CNN for the classification of chest X-ray images. The dataset that they have used contains both pneumonia and non-pneumonia images. Their model has produced satisfactory results. (Ohata et al., 2021) have applied the concept of transfer learning for the automatic identification of Covid-19 based on chest X-ray images. They have used different architectures of CNNs trained on ImageNet and then modified them in order to extract the features from the X-ray images. The CNNs have then been combined with other ML algorithms like K-Nearest Neighbours, Bayes, Random Forest, Multi-layer Perceptron (MLP) and Support Vector Machine (SVM). The test results that they have obtained have been quite encouraging. (Karhan and Akal, 2020) have used ResNet50 model, which is a convolutional neural network architecture for the detection of Covid-19 using chest X-ray images. The experimental results have been quite promising. (Asif et al., 2020) have proposed the implementation of deep convolutional neural network (DCNN) based model Inception V3 with transfer learning for the automatic identification of